whisper-large-v7-twi-places-name-lora-finetuned

This model is a fine-tuned version of prince4332/whisper-v3-akan-complete on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6630
• Wer: 100.0

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.0209	1.0	16	6.3860	182.3529
3.8582	2.0	32	3.9062	129.4118
3.3167	3.0	48	3.1882	123.5294
1.8576	4.0	64	2.3018	105.8824
1.45	5.0	80	1.9975	111.7647
0.9055	6.0	96	1.9082	111.7647
1.4427	7.0	112	1.8038	105.8824
1.1517	8.0	128	1.7697	105.8824
0.6253	9.0	144	1.7348	105.8824
0.8943	10.0	160	1.6997	105.8824
0.8082	11.0	176	1.6769	105.8824
0.5727	12.0	192	1.6630	100.0

Framework versions

• PEFT 0.17.1
• Transformers 4.57.0
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1